Laith Wehbi

Bicycle manufacturers face frequent supply disruptions due to long lead times, global component sourcing and complex product structures. When components are shared across product lines, a single delayed shipment can affect dozens of production orders. Practitioners typically respond with ad hoc expediting, yet have little guidance on which intervention is most effective. This paper develops and evaluates a multi-horizon framework for disruption management in bicycle manufacturing comprising three modules: (i) a planning module that adjusts already-open purchase orders 2–8 weeks ahead via mixed-integer programming, (ii) a scheduling module that swaps at-risk and safe production orders within 1–7 days and (iii) a real-time module that decides on expediting or delaying orders within 24 hours using reinforcement learning. All modules share a common risk metric based on component buffer times and are coordinated through a data-driven decision table that recommends which module to apply based on urgency, exposure and schedule flexibility. Using seven months of production data from a single large European bicycle manufacturer, we evaluate the framework through an offline simulation that compares it against a reconstructed no-intervention baseline, a reactive expediting baseline and fixed single-module strategies. In this case-based simulation, the framework reduces cumulative production delay by 38.9% relative to no intervention and improves on-time completion from 67.3% to 79.6%, with a 4.7% increase in inventory holding cost and 41.2% lower intervention spending than the reactive expediting baseline. The results show that a shared risk metric and simple module-selection logic can support disruption response in a bill-of-materials (BOM)-driven manufacturing setting. Much of the benefit in this case-based simulation comes from matching responses to disruption characteristics rather than applying a single fixed intervention. The framework remains beneficial when lead-time assumptions are tightened and when scheduling soft costs are introduced, although the magnitude of the benefit narrows under stricter assumptions. Realising it in practice depends on data readiness, cross-functional governance and recalibration to firm-specific characteristics. Within these boundaries, the results provide evidence of the value of coordinating planning, scheduling and real-time responses through a shared risk signal.

Timely maintenance of heavy equipment is crucial in construction yet hindered by limited adoption of data-driven methods. We introduce an integrated framework that links priority scoring, predictive analytics, and routing optimization. Our system achieves a 64.4% reduction in prediction error versus a naïve current-SMU baseline on a ten-machine evaluation set without specialized monitoring. Our machine learning models attain MAPE = 4.24% (labor cost) and 9.32% (service duration), substantially outperforming company baselines. We model routing as a Multi-Depot Capacitated Vehicle Routing Problem (MDCVRP) and solve it with a modified Cluster-First, Route-Second approach combining ALNS, 2-OPT*, and a novel SOM-based adaptation. Across varying instances, results show a 41.93% reduction in total time and a 43.86% improvement in high-priority route position relative to current practice. Sensitivity analyses (priority weights, clustering, travel-time uncertainty) confirm robustness. The framework supports better availability and cost savings, and can be extended to multi-equipment and stochastic, real-time settings.

We examine the optimization of order fulfillment strategies employed by IKEA for orders placed via its online webshop. IKEA currently utilizes a strategy whereby orders are assigned to local stores or warehouses based on the minimization of picking and delivery costs without consideration for the potential impact of future orders. This approach often leads to the early exhaustion of order picking capacity in several stores, resulting in increased shipping costs and, in some cases, the inability to fulfill orders as orders received later in the day are assigned to more distant stores. We propose alternative strategies aimed at enhancing the robustness of IKEA's order allocation process against incoming future orders. Our solutions include a heuristic approach that introduces a penalty for selecting warehouses with low residual capacity, a linear programming model that integrates predictions of future orders, and a neural network that is trained on optimal (offline) solutions derived from historical order allocation data. We show that these proposed methods can significantly reduce both the total fulfillment costs and the proportion of orders that cannot be fulfilled, offering a more efficient and reliable order fulfillment system for IKEA.

As modern cities expand, parcel carriers are obliged to deal with the various last-mile challenges of the urban setting, in order to meet the ever-increasing requirements of consumers while ensuring an eco-friendly distribution system. Some of the urban last-mile challenges, such as kerbside parking limitations and vehicle access restrictions, pose additional constraints to carriers that operate van-based deliveries. This paper describes a new distribution model and an associated optimisation problem, namely a vehicle routing problem with portering (VRP-P) with time windows, that integrates on-foot porters with cargo vans. In this distribution strategy, vans rendezvous with porters at handover points at selected customer locations where the transfer of parcels can take place, avoiding the need for designated parking spaces. The VRP-P is modelled as a mixed-integer linear programming formulation and solved on real delivery data in central London. The paper also describes a fast heuristic that uses several improvement operators. Computational results show that the use of portering yields time and carbon savings of up to 89%, depending on the number of customers served. The heuristic obtains high-quality feasible solutions in short computational times and outperforms a commercial optimiser on large scale instances. The findings also show high utilisation of portering even at limited parcel carrying capacities.

E-commerce retailers usually assign each order to the lowest-cost available warehouse, which exhausts capacity in busy periods and drives fulfillment costs up. This paper develops and tests two alternatives, a capacity-aware heuristic with an adaptive busy-day factor and a repeated integer program over sampled future demand, on real data from a large retailer. Both methods outperform the myopic baseline, and the heuristic delivers most of the integer program's gains at a fraction of the compute.